The Visual Guide to Incorporating AI into Today's Classrooms

Synchronizing Graphic Interpretations of Student-AI Engagement

My Cartography Teachers

I've always had a deep affection for charts, maps, diagrams, models, and dioramas. These tools manage to distill complex systems into discernible patterns, enabling us to navigate and understand our world in unique ways.

1. Jorge Luis Borges

After I graduated from college, I decided to move to California. My best friend’s father offered to drive me and my possessions out to the West Coast for a nominal fee. In the backseat, I passed the time reading the short fiction of Jorge Luis Borges, and to my delight came across his amazing parable and mediation about World Maps.  

Borges’ photographers always seem to catch him mid-sentence.

Borges's short narrative tells the story of a cartographer obsessed with creating a map so detailed, it mirrored the entire earth's surface. There I was, at the threshold of a new chapter on the other side of the continent, contemplating the irony of desiring such a map. I have always been one for planning every step of a journey well in advance, yet Borges's narrative made me ponder: in possessing such a detailed map, would one risk missing the essence of the journey, and with it, the world entirely?

2. Paul Feyerabend

My Californian chapter began with odd jobs, eventually leading me to a part-time role at a theological library. In the basement, there were shelves of uncatalogued books. Late one night, I came across historian of science Paul Feyerabend’s Against Method. His heated critique of the scientific method, challenging the foundations of renowned charts and models from giants like Copernicus, Galileo, and Newton (“empirically ungrounded”), was both alarming and inciting. Feyerabend’s relentless questioning and bold assertions made me feel a little nauseous, challenging my perceptions and leaving a lasting impact.

Feyerabend appears to be pondering the irrationality of hair.

3. Bruno Latour

Years later, while studying at Ohio State University, I had another significant encounter with the world of cartography and chart-making. It happened in an unexpected setting — a course on Comics and Illness Narratives. A trusted colleague shared a strange article by Bruno Latour, focusing on the integral role of models in the scientific revolution. Latour’s perspective was a revelation. He wrote about models with reverence, ascribing to them a kind of human-like agency within the dialectic of scientific discovery. 

Latour exists here tangentially in relation to a network of city-objects.

For Latour, and increasingly for me, charts, maps, diagrams, models, and dioramas weren’t just tools; they were active participants in our quest for knowledge, possessing power to guide practitioners to unseen pathways and perspectives. This reinforced a long-held belief of mine, a belief that these objects were more than mere representations; they were key actors in our continuous dialogue with the world.

4 Ways to Map the Future of Education and AI

1. The Decision Tree

In today’s newsletters, I am showcasing how educators and researchers are visualizing the future of AI and education. As my prelude suggests, these charts play important functions beyond gathering together talking points, organizing concepts, or mapping out possible workflows. When employed in a comparative and collaborative manner, these tools acquire an agent-like role, assisting in the development of the AI-responsive writing curriculum that both we and our students seek

"Guide for Students: Should I Use AI?" presents a decision-tree chart by Amanda Bickerstaff, outlining three distinct pathways for AI use: "Yes," "Yes, but," and "No."

"Yes" suggests AI is beneficial for three key tasks: brainstorming, simplifying, and editing.

"Yes, but" advises cautious use of AI in research. AI for Education suggests using Perplexity.ai with built-in online access to lower inaccuracies. Readers should also know that Bing now offers free access to ChatGPT4 that also has built-in online access. 

"No" recommends against using AI to entirely complete an assignment.

This diagram serves as a cognitive map for both teachers and students, guiding their engagement with AI. The degree to which students embrace and apply this framework will be indicative of the effectiveness of AI literacy and AI-responsive writing curricula. Bickerstaff’s approach is notable for its straightforwardness and its potential to empower students. I advocate for educators to actively incorporate this decision-tree into both their professional practices and student learning processes.  

2. The Spiraling Coil

Jessica L. Parker, the founder of Gen AI in Academia Advocate, conceptualizes an AI-responsive writing curriculum as a spiraling coil. Her infographic, titled “Hybrid Human-AI Writing Process,” portrays the collaboration between humans and AI as a fluid, ongoing interaction, representative of her organization's higher education focus. At the heart of this model is the human-AI partnership, driving modern writing practices. The process begins when a student receives an assignment and progresses through recursive stages: (1) brainstorming, (2) drafting, (3) receiving immediate feedback, (4) negotiating meaning, and (5) revising. 

Intentionally, Parker's model allows for either human or AI to contribute at any stage of the writing process, with the underlying expectation that the student remains the decisive authority over the final output. The most significant innovation in Parker’s approach is the concept and stage of 'meaning negotiation.' Fundamentally, I’d argue that all AI-assisted writing can be viewed as a negotiation of meaning. This negotiation, somewhat ironically, occurs with tools that, currently, do not fully grasp meaning in the human sense. However, rather than viewing this as a limitation, models like Parker’s frame it as a chance to situate AI alongside many other writing tools that help us negotiate meaning. The value of these new technologies lie in their ability to help us explore new, engaging, and meaningful directions in our discursive environments, qualifying it as a collective benefit. The real challenge remains how to teach students to effectively negotiate meaning with AI. Here, Parker’s model emerges as a promising start, particularly apt for college-level students and higher.

3. The Parallel Circuit

In my model, I adopt a more segmented approach, aiming to distinctly categorize the varied types of work integral to an AI-responsive writing curriculum. This model is primarily tailored for middle and high school settings, though it's worth noting that Lance Cummings and Marc Watkins have successfully applied a similarly segmented approach in first- and second-year college classrooms. On the left side of my model, I outline the traditional skills of a conventional writing curriculum, which form the bedrock for students to effectively engage with AI, a process that, in turn, further enhances these foundational skills. On the right side, I detail the new competencies necessary for optimizing large language models. These include basic AI literacy, prompt engineering, AI search techniques, conversational skills with AI, and understanding AI ethics.

In practice, these skills are taught concurrently in more advanced grades, yet there’s merit in conceptually visualizing these paths as distinct, given the unique nature of each skill set. At its core, this model illustrates the merging of knowledge-driven and information-driven processes within the act of writing. It underscores that writing decisions often run on parallel and sometimes conflicting paths, and resolving these often necessitates referring to a higher level of purpose, criteria, or objective to produce a satisfactory outcome.

4. The Double Loop

Earlier this week, I received a link from Elliot Bendoly of Ohio State’s Fisher Business School to his latest work, "The role of generative design and additive manufacturing capabilities in developing human–AI symbiosis: Evidence from multiple case studies." In this publication, Bendoly and his team examine the influence of generative design technology on engineering practices during the pandemic, presenting strong evidence for the effective use of these tools.

The study's Analysis and Interpretation section introduces the concept of a "double-loop learning cycle" to describe engineers' interactions with generative technology. Bendoly and colleagues note that these cycles have been discussed in educational and industrial design contexts since the 1960s and 1970s.

In the first loop of their model, designers engage with real-world systems, gather feedback, and make informed decisions. The second loop of Bendoly et al.'s model outlines two trajectories, based on whether the designer opts to employ a generative process. Intriguingly, they describe the use of generative design technology as a transformation in designers' mental models regarding the design process and context.

While there are parallels between this model and earlier ones, the concept of a nested spiral or double loop in Bendoly et al.’s work adds a nuanced layer of understanding. This refinement is particularly valuable as we develop and structure AI-responsive writing curricula and instructional strategies.

Conclusion

Charts, maps, models, and diagrams are invaluable during this era of radical educational transformation. Each of these diagrams unfurls distinct visions for student engagement with generative AI, as they navigate their academic tasks and gear up for future careers. Rather than perceiving these visions as conflicting, I encourage my readers to embrace the splendid diversity they represent. This multifaceted endeavor requires collective effort to craft effective, engaging, and ethical solutions for our students.

Bickerstaff's model offers immediate, practical guidance for educators: discerning when and when not to utilize AI. Parker, on the other hand, gifts us with the profound concept of "negotiating meaning," a cornerstone around which a comprehensive curriculum can be developed. Alongside Lance Cummings and Marc Watkins, I aim to illuminate the specific skills and knowledge essential for students to adeptly negotiate meaning. Lastly, Bendoly et al. connect us to a scholarly discourse predating Large Language Models, reminding us that despite the novelty of our times, there exists a wealth of resources to enrich the curriculum we forge for our students.

Thanks again for reading Educating AI!

Nick Potkalitsky, Ph.D.